Auto-sync: 2026-06-30 10:31:48 (part 2)

results/paper_analysis.json

@@ -1,6 +1,6 @@
 {
   "best_clean_key": "residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmchallenger001",
-  "generated_utc": "2026-06-30T14:17:14+00:00",
+  "generated_utc": "2026-06-30T14:36:44+00:00",
   "mechanism_gap": {
     "best_clean_vs_direct_same_ckpt": 0.0776811594202898,
     "best_clean_vs_h16": 0.06318840579710144,
@@ -8131,9 +8131,73 @@
       "std_success": 0.04906690775535958
     },
     "typed_proposal_lattice_nooponly_bestpolicy_prepend_margin005": {
+      "ci95_success": 0.09239730462675538,
       "label": "Typed proposal lattice head, no-op family only, best-policy checkpoint, policy-prepended margin 0.05",
-      "missing": true,
-      "source": "results/h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_bestpolicy_nooponly_prepend_margin0p05_summary.json"
+      "mean_action_mse_to_best": 0.4323545880660506,
+      "mean_progress": 0.5653533038507769,
+      "mean_success": 0.33159420289855074,
+      "num_completed": 3,
+      "per_task_success": {
+        "LiftPegUpright-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.2673620614501921,
+          "std_success": 0.07514330008398203
+        },
+        "PickCube-v1": {
+          "mean_num_groups": 196.66666666666666,
+          "mean_success": 0.3076465603639517,
+          "std_success": 0.06567867497222973
+        },
+        "PullCube-v1": {
+          "mean_num_groups": 81.0,
+          "mean_success": 0.15722766512240197,
+          "std_success": 0.03551848640237177
+        },
+        "PushCube-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.6901779597166158,
+          "std_success": 0.059178430141350116
+        },
+        "StackCube-v1": {
+          "mean_num_groups": 93.33333333333333,
+          "mean_success": 0.2107966107966108,
+          "std_success": 0.09966298208005851
+        }
+      },
+      "seed_action_mse_to_best": {
+        "0": 0.40204733323971437,
+        "1": 0.4281022290192018,
+        "2": 0.46691420193923555
+      },
+      "seed_progress": {
+        "0": 0.5511523539054652,
+        "1": 0.5908677209994715,
+        "2": 0.5540398366473939
+      },
+      "seed_success": {
+        "0": 0.2991304347826087,
+        "1": 0.37217391304347824,
+        "2": 0.3234782608695652
+      },
+      "selected_candidate_type_counts": {
+        "policy_continuous": 1511,
+        "proposal_no_op": 214
+      },
+      "selected_residual_scale_counts": {},
+      "selected_type_outcomes": {
+        "policy_continuous": {
+          "count": 1511.0,
+          "mean_progress": 0.5792552743250077,
+          "success_rate": 0.3368630046326936
+        },
+        "proposal_no_op": {
+          "count": 214.0,
+          "mean_progress": 0.4671949983060917,
+          "success_rate": 0.29439252336448596
+        }
+      },
+      "source": "results/h16_bestpolicy_nooponly_prepend_margin0p05_summary.json",
+      "std_success": 0.03719191868369956
     },
     "typed_proposal_lattice_nooponly_prepend_margin005": {
       "ci95_success": 0.11324557534724171,
@@ -8204,10 +8268,85 @@
       "source": "results/h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_bestpt_proposal_lattice_nooponly_prepend_margin0p05_summary.json",
       "std_success": 0.04558379972897795
     },
+    "typed_proposal_lattice_nooponly_sparsehead_prepend_margin005": {
+      "label": "Typed proposal lattice head, no-op-only head retrain, policy-prepended margin 0.05",
+      "missing": true,
+      "source": "results/h16_typedprop_nooponly_sparsehead_prepend_margin0p05_summary.json"
+    },
     "typed_proposal_lattice_types2sparse_bestpolicy_prepend_margin005": {
+      "ci95_success": 0.09107461025155232,
       "label": "Typed proposal lattice head, sparse families, best-policy checkpoint, policy-prepended margin 0.05",
-      "missing": true,
-      "source": "results/h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_bestpolicy_types2sparse_prepend_margin0p05_summary.json"
+      "mean_action_mse_to_best": 0.4511233652255116,
+      "mean_progress": 0.5570779382899079,
+      "mean_success": 0.32289855072463763,
+      "num_completed": 3,
+      "per_task_success": {
+        "LiftPegUpright-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.25212622530385503,
+          "std_success": 0.07912705997645623
+        },
+        "PickCube-v1": {
+          "mean_num_groups": 196.66666666666666,
+          "mean_success": 0.3076465603639517,
+          "std_success": 0.06567867497222973
+        },
+        "PullCube-v1": {
+          "mean_num_groups": 81.0,
+          "mean_success": 0.1564884433305486,
+          "std_success": 0.014120874822966888
+        },
+        "PushCube-v1": {
+          "mean_num_groups": 102.0,
+          "mean_success": 0.655894255325475,
+          "std_success": 0.09882691423366702
+        },
+        "StackCube-v1": {
+          "mean_num_groups": 93.33333333333333,
+          "mean_success": 0.2107966107966108,
+          "std_success": 0.09966298208005851
+        }
+      },
+      "seed_action_mse_to_best": {
+        "0": 0.42280645585758103,
+        "1": 0.44845469078160416,
+        "2": 0.48210894903734974
+      },
+      "seed_progress": {
+        "0": 0.5449448479031739,
+        "1": 0.5814847025639661,
+        "2": 0.5448042644025839
+      },
+      "seed_success": {
+        "0": 0.2921739130434783,
+        "1": 0.3634782608695652,
+        "2": 0.3130434782608696
+      },
+      "selected_candidate_type_counts": {
+        "policy_continuous": 1428,
+        "proposal_no_op": 197,
+        "proposal_wrong_gripper": 100
+      },
+      "selected_residual_scale_counts": {},
+      "selected_type_outcomes": {
+        "policy_continuous": {
+          "count": 1428.0,
+          "mean_progress": 0.5722373958236175,
+          "success_rate": 0.32072829131652664
+        },
+        "proposal_no_op": {
+          "count": 197.0,
+          "mean_progress": 0.46950423573781996,
+          "success_rate": 0.29441624365482233
+        },
+        "proposal_wrong_gripper": {
+          "count": 100.0,
+          "mean_progress": 0.5131210787361488,
+          "success_rate": 0.41
+        }
+      },
+      "source": "results/h16_bestpolicy_types2sparse_prepend_margin0p05_summary.json",
+      "std_success": 0.036659505515970706
     },
     "typed_proposal_lattice_types2sparse_prepend_margin005": {
       "ci95_success": 0.10288081677597381,

results/paper_analysis.md

@@ -1,6 +1,6 @@
 # Paper Analysis
 
-Generated: `2026-06-30T14:17:14+00:00`
+Generated: `2026-06-30T14:36:44+00:00`
 
 ## Main Seed Statistics
 
@@ -65,8 +65,9 @@ Generated: `2026-06-30T14:17:14+00:00`
 | typed_proposal_lattice_types4safe_prepend_margin005 | Typed proposal lattice head, safe four families, policy-prepended margin 0.05 | 3 | 33.45% +/- 3.39 | +/- 8.43 | 55.47% | 0.456 | +3.71 pp |
 | typed_proposal_lattice_types2sparse_prepend_margin005 | Typed proposal lattice head, sparse no-op/wrong-gripper families, policy-prepended margin 0.05 | 3 | 34.26% +/- 4.14 | +/- 10.29 | 56.08% | 0.461 | +4.52 pp |
 | typed_proposal_lattice_nooponly_prepend_margin005 | Typed proposal lattice head, no-op family only, policy-prepended margin 0.05 | 3 | 34.43% +/- 4.56 | +/- 11.32 | 56.23% | 0.450 | +4.70 pp |
-| typed_proposal_lattice_types2sparse_bestpolicy_prepend_margin005 | Typed proposal lattice head, sparse families, best-policy checkpoint, policy-prepended margin 0.05 | 0 | missing | missing | missing | missing | missing |
-| typed_proposal_lattice_nooponly_bestpolicy_prepend_margin005 | Typed proposal lattice head, no-op family only, best-policy checkpoint, policy-prepended margin 0.05 | 0 | missing | missing | missing | missing | missing |
+| typed_proposal_lattice_types2sparse_bestpolicy_prepend_margin005 | Typed proposal lattice head, sparse families, best-policy checkpoint, policy-prepended margin 0.05 | 3 | 32.29% +/- 3.67 | +/- 9.11 | 55.71% | 0.451 | +2.55 pp |
+| typed_proposal_lattice_nooponly_bestpolicy_prepend_margin005 | Typed proposal lattice head, no-op family only, best-policy checkpoint, policy-prepended margin 0.05 | 3 | 33.16% +/- 3.72 | +/- 9.24 | 56.54% | 0.432 | +3.42 pp |
+| typed_proposal_lattice_nooponly_sparsehead_prepend_margin005 | Typed proposal lattice head, no-op-only head retrain, policy-prepended margin 0.05 | 0 | missing | missing | missing | missing | missing |
 | residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmbonus001 | K4 composed compatible tangents, no-op bonus 0.03, singleton near-miss bonus 0.01 | 3 | 35.59% +/- 0.99 | +/- 2.46 | 57.10% | 0.406 | +5.86 pp |
 | residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmbonus002 | K4 composed compatible tangents, no-op bonus 0.03, singleton near-miss bonus 0.02 | 3 | 35.59% +/- 0.99 | +/- 2.46 | 57.10% | 0.406 | +5.86 pp |
 | residual_k4_composemasked_dropnmnoop_grid035040045_srcscorebonus002 | K4 composed type-consensus tangents, masked, drop near-miss+no-op composite, source-score bonus 0.02 | 3 | 35.48% +/- 1.22 | +/- 3.02 | 57.02% | 0.408 | +5.74 pp |

results/paper_core_results.md

@@ -61,7 +61,7 @@ lattice is `+27.25 pp`, and the remaining clean-to-same-state proposal gap is
 | K4 compatible residual retrieval, unique candidate-oracle prefix K=8 | No | No | 43.07% diagnostic | +13.33 pp diagnostic | Diagnostic-only measured oracle over generated clean candidate prefix; unique-action trace shows real selector headroom, not duplicate-action inflation |
 | K4 compatible residual retrieval + near-miss challenger gate 0.01 | No | No | 36.06% | +6.32 pp | Current best clean deployment row; a two-stage selector keeps the robust compatible-chart anchor and only lets singleton near-miss tangents override under a tightly calibrated positive field margin |
 | K4 compatible residual retrieval + near-miss challenger fine calibration | No | No | 36.00-36.00% | +6.26 pp | Margins 0.005/0.015/0.02 form a near-tie plateau around the 0.01 optimum; scale-gating the challenger to 0.35 or 0.35/0.40 lowers MSE slightly but does not recover the lost success; 0.03 and near-miss+wrong-gripper are lower or higher-MSE, so the effect is calibrated singleton near-miss geometry rather than broad residual mixing |
-| Typed proposal lattice head, generated primitive families | No | No | 31.30-34.43% | +1.57-4.70 pp | Direct generated support improves as anti-goal families are masked: six-family is 31.30%/32.17%, safe4 is 33.45%, sparse no-op/wrong-gripper is 34.26%, and no-op-only is 34.43%; still below compatible tangent transport, so proposal locality remains the bottleneck. Best-policy checkpoint ablations are pending as `14963143`/`14963147` |
+| Typed proposal lattice head, generated primitive families | No | No | 31.30-34.43% | +1.57-4.70 pp | Direct generated support improves as anti-goal families are masked: six-family is 31.30%/32.17%, safe4 is 33.45%, sparse no-op/wrong-gripper is 34.26%, and no-op-only is 34.43%; best-policy checkpoint ablations drop to 32.29%/33.16%, and a no-op-only sparse-head retrain is pending |
 | K4 compatible residual retrieval, margin 0.10 | No | No | 34.67% | +4.93 pp | Naively lowering abstention selects too many bad nonzero tangents and falls below the 36.06% top row |
 | K4 composed type-consensus residual retrieval, exact compatibility mask + source-score prior | No | No | 35.48% | +5.74 pp | Measured train-source reward confidence does not replace the sparse typed prior on the compatible chart; useful negative calibration |
 | K4 composed type-consensus residual retrieval, exact compatibility mask + no-op/source-score priors | No | No | 35.54% | +5.80 pp | Adding measured train-source reward confidence to the current typed-prior chart ties near-best but does not beat exact compatibility masking alone |

results/paper_story_memo.md

@@ -34,7 +34,7 @@ when queried on proposal geometry that matches those local counterfactuals.
 | Singleton near-miss priors do not open a new clean route | adding a small exact `residual_near_miss` singleton prior 0.01 or 0.02 on the compatible chart ties the previous 35.59% compatibility row and slightly raises progress to 57.10% | Tie diagnostic: revived near-miss tangents are high-precision but too sparse to close the clean-to-same-state proposal gap |
 | Candidate-prefix oracle needs unique-action hygiene | the first K=8 prefix diagnostic was archived as `_nonunique`; the deduplicated unique-action trace reaches 43.07% candidate-oracle success with mean best branch rank 2.85 | Supported diagnostic: proposal headroom is real, but branch success falls with rank, so selector calibration must be conditional |
 | Near-miss challenger calibration improves clean deployment | the trace-motivated two-stage selector keeps the compatible-chart anchor and lets singleton near-miss tangents override only under a tightly calibrated margin; margin 0.01 reaches 36.06%, while 0.005/0.015/0.02 and scale-gated 0.35 or 0.35/0.40 variants tie at 36.00% | Current best clean row; small but story-aligned selector-calibration gain, with scale-gating explaining MSE but not improving success and wrong-gripper challengers rejected by lower success or higher MSE |
-| Typed proposal generation confirms support geometry, not just support count, is the remaining clean bottleneck | six-family generated typed proposal lattices reach 31.30%/32.17%; masking anti-goal families raises safe4 to 33.45%, sparse no-op/wrong-gripper to 34.26%, and no-op-only to 34.43%, still below the 36.06% compatible-tangent row | Negative/diagnostic: direct proposal generation helps only when sparse and anti-goal masked, but learned proposals still lack the local same-state counterfactual geometry |
+| Typed proposal generation confirms support geometry, not just support count, is the remaining clean bottleneck | six-family generated typed proposal lattices reach 31.30%/32.17%; masking anti-goal families raises safe4 to 33.45%, sparse no-op/wrong-gripper to 34.26%, and no-op-only to 34.43%; best-policy checkpoint ablations drop to 32.29%/33.16%; no-op-only sparse-head retrain is pending | Negative/diagnostic: direct proposal generation helps only when sparse and anti-goal masked, but learned proposals still lack the local same-state counterfactual geometry |
 | Component-wise composite priors do not add the gain | propagating the no-op bonus into composite types reaches 35.36%, below the exact-prior masked composition row at 35.54% and the exact compatibility row at 35.59% | Negative/near-tie diagnostic: sparse exact priors are better than broadly rewarding every no-op-containing composite |
 | Composite trust-radius penalty explains but does not improve the success top line | composite-only L2 penalty 0.02 ties 35.54% while lowering action MSE from 0.4106 to 0.4079; with the exact compatibility mask it lowers MSE further to 0.4048 but drops back to 35.54% | Tie/negative diagnostic: composed tangents need a local trust radius, but compatibility masking gives the success optimum |
 | Minimum-energy residual regularization does not add the gain | action L2 penalty 0.05 ties the previous 35.42% scale-grid row, while 0.10/0.20 reach 35.36% | Negative/tie diagnostic: the clean bridge is not explained by shortest-action bias |
@@ -85,7 +85,7 @@ clean proposal result, the intended main rows are:
 17. K4 masked composed type-consensus tangent transport: 35.30%; with exact near-miss/no-op compatibility mask only: 35.48%; with typed no-op prior: 35.54%; with exact compatibility mask + typed no-op prior: 35.59%; adding singleton near-miss prior 0.01/0.02 ties 35.59%; with source-score prior on that exact chart: 35.48%; with typed no-op + source-score: 35.54%
 18. K4 compatible tangent unique-action candidate-oracle prefix K=8 diagnostic: 43.07%; branch trace shows mean best branch rank 2.85, unique count 8.00, and real but conditional selector headroom
 19. K4 compatible tangent near-miss challenger gate 0.01: 36.06%; current best clean deployment row from trace-motivated two-stage selector calibration. Fine margins 0.005/0.015/0.02 tie at 36.00%; scale-gating the challenger to 0.35 or 0.35/0.40 also ties 36.00% with slightly lower MSE; 0.03 and near-miss+wrong-gripper variants are lower or higher-MSE.
-20. Typed proposal lattice head: six-family generated proposals reach 31.30%/32.17%; safe4 reaches 33.45%; sparse no-op/wrong-gripper reaches 34.26%; no-op-only reaches 34.43%. Checkpointing ablations with `best_policy.pt` are pending as `14963143`/`14963147`.
+20. Typed proposal lattice head: six-family generated proposals reach 31.30%/32.17%; safe4 reaches 33.45%; sparse no-op/wrong-gripper reaches 34.26%; no-op-only reaches 34.43%. Checkpointing ablations with `best_policy.pt` are lower at 32.29%/33.16%. A no-op-only sparse-head retrain is pending as `14963500` -> `14963505` -> `14963507`.
 21. K4 compatible tangent abstention margin 0.10 diagnostic: 34.67%; naive lower abstention hurts despite candidate-prefix oracle headroom
 22. K4 masked composed type-consensus tangent transport + component-wise no-op prior: 35.36%
 21. K4 masked composed type-consensus tangent transport + composite L2 trust penalty: 35.54% at 0.02; 35.48% at 0.05; 35.54% with exact compatibility mask + L2 0.02

results/paper_table_status.json

@@ -1641,32 +1641,51 @@
     {
       "key": "typed_proposal_lattice_types2sparse_bestpolicy_prepend_margin005",
       "label": "Typed proposal lattice head, sparse no-op/wrong-gripper families, best-policy checkpoint",
-      "path": "h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_bestpolicy_types2sparse_prepend_margin0p05_summary.json",
+      "path": "h16_bestpolicy_types2sparse_prepend_margin0p05_summary.json",
       "clean_deployment": "yes",
       "same_state_proposals": "no",
       "expert_proposal": "no",
       "story_role": "checkpointing ablation for typed generated sparse proposals: lower BC/proposal loss versus weaker field scorer",
       "fallback_success": null,
       "pending_job": "14963143/14963150",
-      "path_exists": false,
-      "status": "pending",
-      "success": null,
-      "std_success": null,
+      "path_exists": true,
+      "status": "complete",
+      "success": 0.3228985507246377,
+      "std_success": 0.036659505515970706,
       "completed_seeds": null,
-      "num_completed": null,
+      "num_completed": 3,
       "best_config": null,
-      "gain_vs_h16_policy": null
+      "gain_vs_h16_policy": 0.02550724637681162
     },
     {
       "key": "typed_proposal_lattice_nooponly_bestpolicy_prepend_margin005",
       "label": "Typed proposal lattice head, no-op family only, best-policy checkpoint",
-      "path": "h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_bestpolicy_nooponly_prepend_margin0p05_summary.json",
+      "path": "h16_bestpolicy_nooponly_prepend_margin0p05_summary.json",
       "clean_deployment": "yes",
       "same_state_proposals": "no",
       "expert_proposal": "no",
       "story_role": "minimal generated no-op checkpointing ablation for typed proposal support",
       "fallback_success": null,
       "pending_job": "14963147/14963149",
+      "path_exists": true,
+      "status": "complete",
+      "success": 0.33159420289855074,
+      "std_success": 0.03719191868369956,
+      "completed_seeds": null,
+      "num_completed": 3,
+      "best_config": null,
+      "gain_vs_h16_policy": 0.03420289855072467
+    },
+    {
+      "key": "typed_proposal_lattice_nooponly_sparsehead_prepend_margin005",
+      "label": "Typed proposal lattice head, no-op-only head retrain",
+      "path": "h16_typedprop_nooponly_sparsehead_prepend_margin0p05_summary.json",
+      "clean_deployment": "yes",
+      "same_state_proposals": "no",
+      "expert_proposal": "no",
+      "story_role": "sparse-head retrain test for whether typed proposal failure comes from multi-family interference",
+      "fallback_success": null,
+      "pending_job": "14963500/14963505/14963507",
       "path_exists": false,
       "status": "pending",
       "success": null,

results/paper_table_status.md

@@ -89,8 +89,9 @@ Baseline h=16 policy: 29.74%
 | typed_proposal_lattice_types4safe_prepend_margin005 | Typed proposal lattice head, safe four families, policy-prepended margin 0.05 | complete | 33.45% | +3.71 pp | yes | no | no | anti-goal proposal-family mask test for the model-generated typed lattice |
 | typed_proposal_lattice_types2sparse_prepend_margin005 | Typed proposal lattice head, sparse no-op/wrong-gripper families, policy-prepended margin 0.05 | complete | 34.26% | +4.52 pp | yes | no | no | sparse safe-support test for whether typed generated no-op/wrong-gripper proposals recover the clean support gap |
 | typed_proposal_lattice_nooponly_prepend_margin005 | Typed proposal lattice head, no-op family only, policy-prepended margin 0.05 | complete | 34.43% | +4.70 pp | yes | no | no | minimal generated no-op proposal-family test after sparse typed lattice improved over the full six-family row |
-| typed_proposal_lattice_types2sparse_bestpolicy_prepend_margin005 | Typed proposal lattice head, sparse no-op/wrong-gripper families, best-policy checkpoint | pending 14963143/14963150 | pending | pending | yes | no | no | checkpointing ablation for typed generated sparse proposals: lower BC/proposal loss versus weaker field scorer |
-| typed_proposal_lattice_nooponly_bestpolicy_prepend_margin005 | Typed proposal lattice head, no-op family only, best-policy checkpoint | pending 14963147/14963149 | pending | pending | yes | no | no | minimal generated no-op checkpointing ablation for typed proposal support |
+| typed_proposal_lattice_types2sparse_bestpolicy_prepend_margin005 | Typed proposal lattice head, sparse no-op/wrong-gripper families, best-policy checkpoint | complete | 32.29% | +2.55 pp | yes | no | no | checkpointing ablation for typed generated sparse proposals: lower BC/proposal loss versus weaker field scorer |
+| typed_proposal_lattice_nooponly_bestpolicy_prepend_margin005 | Typed proposal lattice head, no-op family only, best-policy checkpoint | complete | 33.16% | +3.42 pp | yes | no | no | minimal generated no-op checkpointing ablation for typed proposal support |
+| typed_proposal_lattice_nooponly_sparsehead_prepend_margin005 | Typed proposal lattice head, no-op-only head retrain | pending 14963500/14963505/14963507 | pending | pending | yes | no | no | sparse-head retrain test for whether typed proposal failure comes from multi-family interference |
 | retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmbonus001 | K4 composed compatible residual retrieval, no-op bonus 0.03, singleton near-miss bonus 0.01 | complete | 35.59% | +5.86 pp | yes | no | no | revive high-precision singleton near-miss tangents without boosting toxic near-miss+no-op composites |
 | retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmbonus002 | K4 composed compatible residual retrieval, no-op bonus 0.03, singleton near-miss bonus 0.02 | complete | 35.59% | +5.86 pp | yes | no | no | stronger singleton near-miss revival prior on the compatible local tangent chart |
 | retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_srcscorebonus002 | K4 composed type-consensus residual retrieval, masked, drop near-miss+no-op composite, source-score bonus 0.02 | complete | 35.48% | +5.74 pp | yes | no | no | train-measured source-score prior on the compatible local tangent chart |

scripts/build_paper_analysis.py

@@ -500,18 +500,17 @@ METHODS = [
     MethodSpec(
         key="typed_proposal_lattice_types2sparse_bestpolicy_prepend_margin005",
         label="Typed proposal lattice head, sparse families, best-policy checkpoint, policy-prepended margin 0.05",
-        summary_path=(
-            "h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_"
-            "bestpolicy_types2sparse_prepend_margin0p05_summary.json"
-        ),
+        summary_path="h16_bestpolicy_types2sparse_prepend_margin0p05_summary.json",
     ),
     MethodSpec(
         key="typed_proposal_lattice_nooponly_bestpolicy_prepend_margin005",
         label="Typed proposal lattice head, no-op family only, best-policy checkpoint, policy-prepended margin 0.05",
-        summary_path=(
-            "h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_"
-            "bestpolicy_nooponly_prepend_margin0p05_summary.json"
-        ),
+        summary_path="h16_bestpolicy_nooponly_prepend_margin0p05_summary.json",
+    ),
+    MethodSpec(
+        key="typed_proposal_lattice_nooponly_sparsehead_prepend_margin005",
+        label="Typed proposal lattice head, no-op-only head retrain, policy-prepended margin 0.05",
+        summary_path="h16_typedprop_nooponly_sparsehead_prepend_margin0p05_summary.json",
     ),
     MethodSpec(
         key="residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmbonus001",

scripts/build_paper_table_status.py

@@ -878,7 +878,7 @@ SPECS = [
     ResultSpec(
         key="typed_proposal_lattice_types2sparse_bestpolicy_prepend_margin005",
         label="Typed proposal lattice head, sparse no-op/wrong-gripper families, best-policy checkpoint",
-        path="h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_bestpolicy_types2sparse_prepend_margin0p05_summary.json",
+        path="h16_bestpolicy_types2sparse_prepend_margin0p05_summary.json",
         clean_deployment="yes",
         same_state_proposals="no",
         expert_proposal="no",
@@ -888,13 +888,23 @@ SPECS = [
     ResultSpec(
         key="typed_proposal_lattice_nooponly_bestpolicy_prepend_margin005",
         label="Typed proposal lattice head, no-op family only, best-policy checkpoint",
-        path="h16_policy_ckpt_near_miss_policy_bc5_typedprop_p2_bestpolicy_nooponly_prepend_margin0p05_summary.json",
+        path="h16_bestpolicy_nooponly_prepend_margin0p05_summary.json",
         clean_deployment="yes",
         same_state_proposals="no",
         expert_proposal="no",
         story_role="minimal generated no-op checkpointing ablation for typed proposal support",
         pending_job="14963147/14963149",
     ),
+    ResultSpec(
+        key="typed_proposal_lattice_nooponly_sparsehead_prepend_margin005",
+        label="Typed proposal lattice head, no-op-only head retrain",
+        path="h16_typedprop_nooponly_sparsehead_prepend_margin0p05_summary.json",
+        clean_deployment="yes",
+        same_state_proposals="no",
+        expert_proposal="no",
+        story_role="sparse-head retrain test for whether typed proposal failure comes from multi-family interference",
+        pending_job="14963500/14963505/14963507",
+    ),
     ResultSpec(
         key="retrieval_residual_k4_composemasked_dropnmnoop_grid035040045_noopbonus003_nmbonus001",
         label="K4 composed compatible residual retrieval, no-op bonus 0.03, singleton near-miss bonus 0.01",